GPT-Image-2.6 Flare vs Sunburst: Which One to Use, and What's New
OpenAI released two new image models on 4 November: Flare for quick drafts, Sunburst for detailed final work. Both can now make SVG files and short GIFs. Prices, differences and a simple way to use both.
यह लेख हिंदी में पढ़ेंGPT-Image-2.6 Flare vs Sunburst: Which One to Use, and What's New
OpenAI added two image models to its API on 4 November: GPT-Image-2.6 Flare and GPT-Image-2.6 Sunburst. OpenAI's one-line descriptions are the clearest way to tell them apart. Flare is for 'fast iteration'. Sunburst is for 'detailed visual work'. They have the same features and the same price, so the choice is only about speed against detail.
What is new in 2.6
- SVG output. You can ask for a vector file instead of a PNG. That matters for logos, icons and anything that has to print sharp at any size. It works when generating, not when editing.
- Animated GIFs. Short looping animations straight from the Image API, useful for stickers, simple product loops and reaction images.
- Markup edits. Draw on top of an image (circle the thing, write 'make this red') and send the drawing with your request. OpenAI charges nothing extra for it.
- Complexity control. A new setting with low, medium and high, which defaults to high. Low is for simple, clean images; high is for busy scenes with lots of detail.
What it costs
| Per 1M tokens | Flare | Sunburst |
|---|---|---|
| Text input | $5 | $5 |
| Image input | $8 | $8 |
| Image output | $30 | $30 |
OpenAI prices both by tokens, not per picture, so the cost of one image depends on its size and quality setting. The prices are identical, which means the only reason to use Flare is that it is quicker. A third-party test reported Flare at around 6 seconds and Sunburst at around 9 per image. Those numbers are not from OpenAI.
A simple way to use both
Draft on Flare, finish on Sunburst. Try ten versions of a Diwali poster or a product shot on Flare until the layout and colours are right, then run the final prompt on Sunburst for the version you will actually print or post. You save the waiting, not money, since the price is the same either way.
If you need text inside the image, like a shop name or an offer line, use Sunburst for the final version and keep the text short. Long prompts and long text are where faster image models usually slip first.
Is it in ChatGPT
OpenAI's changelog lists it for the API (image generation and image edits) and says nothing about which ChatGPT plans get it. I couldn't find an OpenAI statement either way, so if you only use ChatGPT, don't assume the image tool you see there is 2.6. If you want to try it without code, tools that connect to OpenAI's API will add it first.
Want prompts to try it with? Our image prompt library has ready ones for posters, thumbnails and product shots, each with the picture it made.
Sources
Frequently asked questions
What is the difference between GPT-Image-2.6 Flare and Sunburst?
Flare is built for fast iteration and Sunburst for detailed visual work. They have the same features and the same price, so pick Flare for drafts and Sunburst for the final image.
How much does GPT-Image-2.6 cost?
Both cost $5 per million text input tokens, $8 per million image input tokens and $30 per million image output tokens. There is no fixed price per image.
Can GPT-Image-2.6 make SVG and GIF files?
Yes. Both models support SVG output when generating and short animated GIFs through the Image API, according to OpenAI's guide.
Is GPT-Image-2.6 available in ChatGPT?
OpenAI's announcement covers the API only. We found no statement on which ChatGPT plans have it.
Save this summary as an image or share it.
AICreatorHub Team
The AICreatorHub editorial team is a group of hands-on AI practitioners, writers and developers based in India. We test AI tools and models ourselves, track official releases from OpenAI, Anthropic, Google, Meta and xAI, and translate them into simple, India-first guides in English and Hindi. Every article is written for real Indian use cases — pricing in rupees, free-tier tips and practical, tested steps — so you get accurate, up-to-date and genuinely useful AI information.